How to Govern Account-Level Negative Keywords

An account-level negative keyword is a shared control, not a convenient place to store every poor search term. Google says the account list automatically applies to eligible Search and Shopping inventory in relevant campaign types. That reach makes the list appropriate for expressions that are truly incompatible with the whole business, such as a service, product category, audience, or policy boundary the account never intends to serve. A term that is irrelevant to one product but valuable to another belongs at a narrower level.

Create a short policy before adding entries. Define who can propose a term, which evidence is required, who checks cross-campaign impact, and how quickly a mistaken exclusion can be removed. Inventory brand names, product families, services, geographic language, compatibility terms, and lifecycle queries that appear across the account. Ask whether any active or planned offer could reasonably match the proposed term. If the answer is uncertain, hold the term at campaign or ad-group scope while more evidence is collected.

Build each proposal from real search-term or placement evidence, customer language, policy requirements, or an explicit catalog boundary. Preserve the original query, date range, campaign, match source, cost and conversion definitions, but do not infer intent from one unattributed visit. Normalize casing and obvious duplicates for review while keeping singulars, plurals, synonyms, brands, model names, and qualifiers separate when they change meaning. Google notes that negative keywords do not use close variants in the same way positive keywords do, so a single root may not cover the forms an operator expects.

Assign broad, phrase, or exact negative match deliberately. Negative broad can block a query containing all excluded terms even when their order differs. Phrase requires the excluded phrase in order, and exact is narrower. Write sample queries that should be blocked and should remain eligible before the change. Include long queries, reordered words, abbreviations, and product variants. This test set turns a subjective keyword decision into a checkable contract and helps a reviewer catch exclusions that would remove valuable comparison, support, or branded demand.

List every campaign type and inventory surface that can inherit the account control. Google documents account-level negatives for eligible Search and Shopping inventory and describes their use with Performance Max brand-suitability controls. Do not assume the setting behaves like a campaign-only Search list or that it controls every Display and Video placement in the same way. Treat placement exclusions, content suitability, brand exclusions, audience exclusions, and negative keywords as separate controls with separate evidence.

For each proposed keyword, map the campaigns, feeds, countries, languages, product groups, conversion goals, and launch plans it could affect. Review active and paused campaigns because a paused campaign can return later with inherited settings. Check manager-account practices as well as the child account. If a term is universal only in one country, one business unit, or one product family, use a narrower control. Record the account ID and exact list location so another operator does not recreate the same exclusion elsewhere.

Export the current list before editing. Add a named batch with a ticket or change reference, reviewer, timestamp, match types, expected blocked examples, and expected protected examples. Keep the batch small enough to reverse without reconstructing an entire taxonomy. After saving, reopen the account setting and verify every row. Then use supported ad-preview, diagnostics, and live search-term reporting to confirm the intended effect without generating artificial clicks.

Watch impressions, eligible queries, campaign diagnostics, spend, conversion quality, and zero-delivery conditions across a complete comparison window. A drop is not automatically an improvement, and stable aggregate performance does not prove every product retained coverage. Segment by campaign and product family, then review the protected-query examples. If valuable traffic disappears or a launch campaign cannot serve, remove the responsible entry first and investigate before adding compensating keywords or budgets.

Keep a register containing the normalized term, original evidence, match type, account, affected inventory, reason, proposer, approver, created date, review date, protected queries, and rollback result. Recheck entries when the catalog, geography, service scope, landing pages, policy, campaign mix, or brand strategy changes. A keyword that was universally irrelevant last quarter can become important after a new product launch. Expired exclusions should be removed through the same reviewed process used to create them.

Audit for duplicates, conflicting match types, misspellings, category roots, competitor brands, and terms whose business rationale is no longer visible. Compare the account list with campaign and ad-group lists so overlapping controls do not obscure the cause of lost traffic. Keep platform reporting distinct from business outcomes: fewer unwanted queries can be useful, but it does not by itself prove more profit or stronger demand. Reconcile accepted orders, cancellations, returns, and contribution under the same conversion contract used before the change.

The best account-level list is short enough to explain. Every entry should answer what it blocks, why the whole account should block it, which evidence supported the decision, who approved it, how it was tested, and when it will be reviewed. If those answers cannot be produced, narrow or remove the entry. Governance protects reach while preserving the operational value of a global exclusion.